Wednesday, November 6, 2013

Gigabit Connections Will Be Commonplace by 2020, Really

Predictions always are difficult, under the best of circumstances, because researchers cannot really account for the unexpected, principally unexpected developments that analyst Nassim Taleb refers to as black swan events.

And it now appears all predictions about the typical U.S. Internet access speed, or its pricing, are becoming less reliable as the market becomes more unstable, in what many would consider a very good way.

You might argue Google Fiber, with its symmetrical gigabit service, accelerated the trend. Others might argue that the historic rate of growth of typical Internet access speeds would have suggested gigabit networks would arrive.

The point is that activity leading to higher speeds is happening on a faster scale. Since the Google Fiber launch, many other major ISPs have been committing to, and deploying, higher speed networks. AT&T and CenturyLink are among them, each committing to new gigabit networks in some markets where competition now requires it.

Separately, existing suppliers of gigabit services recently lowered prices from the earlier $300 a month level to match Google Fiber pricing of $70 a month, or come close, at $80 a month. Obviously, Google Fiber now creates the pricing umbrella.

Now the Los Angeles City Council, which has been looking at ways to provide a metro Wi-Fi network providing free service to residents, has decided to explore creation of a new citywide gigabit network.

The Council now apparently will issue a request for proposals for vendors willing to fund and build a fiber to the home reaching every home and business in Los Angeles, with wholesale access requirements as well, to be funded entirely from private sources.

The network would be required to offer free access at rates between 2 Mbps and 5 Mbps, but would be allowed to offer paid service at speeds up to a gigabit per second.

At the same time, new suppliers, seeing new business opportunities in the “heterogeneous networks” trend, are attempting to create new public Wi-Fi networks.

Gowex, a Spanish firm that already provides Wi-Fi networks in more than 80 cities globally, wants to combine public Wi-Fi and private access Wi-Fi services from mobile operators and businesses to create seamless mobile data coverage for consumers in New York, N.Y., and hopes to expand beyond New York later.

Called We-2, the new service will launch in December in New York with an initial network of more than 2000 Wi-Fi hotspots across New York,  with plans to further expand aggressively across the busiest corridors of Manhattan, Queens and The Bronx.

The company also hopes to create We-2 Wi-Fi hotspot networks in more than 300 cities by 2020.

The business model is what makes the initiative different from Fon.

Apparently, We-2 will be a network created and sold to mobile service providers who want Wi-Fi offload capabilities.

The larger point is that most existing predictions about how fast the typical U.S. Internet access connection will be, or what it will cost, are going to be wrong if they do not account for the actual pattern of supply growth we already have seen.

And that pattern suggests growth of two or three orders of magnitude, which would put a typical connection of today (perhaps 15 Mbps) at perhaps a gigabit by 2020 (two orders of magnitude growth in seven years), would simply be in keeping with past trends.

Notably, even large Internet service providers often are unable to accurately forecast how much bandwidth their own networks will require. For example, in a March 2011 presentation AT&T projected that data volumes would grow by eight to 10 times between the end of 2010 and the end of 2015.

That forecast appears to be based on an expectation that volumes would roughly double in 2011 and then increase by a further 65 percent in 2012.

Instead, AT&T in 2012 revised that projection to 40 percent annual growth. Now, 40 percent annual growth is significant. It means bandwidth consumption doubles about every two to three years.  

But annual bandwidth growth of 50 percent a year would be well within historical ranges, on an aggregate basis, in terms of long-haul bandwidth consumption. But policies and end user behavior can change the demand curve.

Some would suggest users learned to shift consumption to Wi-Fi, that lighter users unexpectedly had lighter usage profiles and that end users learned to modify their behavior in ways that reduced overall consumption or demand.

If demand grows at that level (doubling every two to three years), it is obvious that supply also has to grow to match consumption, all other things being equal.

And though it might seem improbable that typical purchased speeds could reach the gigabit level by 2020, that is indeed likely, based strictly on past precedent.

In August 2000, only 4.4 percent of U.S. households had a home broadband connection, while  41.5 percent of households had dial-up access.

A decade later, dial-up subscribers declined to 2.8 percent of households in 2010, and 68.2 percent of households subscribed to broadband service.

In other words, from 2000 to 2012, the typical purchased access connection grew by about two to three orders of magnitude in about a decade.

“Why would a consumer pay for a gigabit connection?” has been a reasonable question, given the costs and expected revenues.
Increasingly, that is the wrong question to ask. The relevant question is “why would a consumer want to buy an Internet access connection?” Speed grows, by about two to three orders of magnitude, every decade.

So “speed” is not the most relevant question. Speed grows. The question is the value of an Internet access connection.

Based on history and demand for access to the Internet, plus the dramatic compression of prices, gigabit connections will be common in 2020.

No comments:

"Tokens" are the New "FLOPS," "MIPS" or "Gbps"

Modern computing has some virtually-universal reference metrics. For Gemini 1.5 and other large language models, tokens are a basic measure...